6-1-analysis-morphind-etc-editing.R

# script to edit morphind tag, wordclass, etc. from within the 6-analysis.R data
verbs <- verbs %>% 
  filter(!word_form %in% c("memperikan", "dipertim", "dipin",
                           "beradaan", # split from "keberadaan"
                           "berbedaan", # typo of "perbedaan"
                           "berbuatan", # typo of perbuatan
                           "berdayaan", # split from pemberdayaan (peN-/-an) and ketidakberdayaan (ke-/-an)
                           "berdin", # typo for berdiri
                           "bergan", # split typo for "bergantung"
                           "beredan", # typo for beredar
                           "bergen", # place name Bergen in Norway
                           "berdoa-menataptersenyum-berjalan", "berkai-kali", "berusahan",
                           "teras-stupa", "terdidik-tercerahkan", "termi-nologi", 
                           "tergan-tung", "terlintas-linlas", "terburu-burumerealisasi",
                           "ber-gantung", "ber-henti", "berkai-kali", "ber-bicara", 
                           "ber-lanjut", "ber-ubah-ubah", "ber-angsur-angsur", 
                           "ber-ulang", "ber-tanya", "bermaian", "bergman", "berjiran", 
                           "memperbaki", "terpen",
                           "berianjak-tanjak", "ber-temu", "ber-juang", 
                           "beri-msiatif", "ber-kumpulnya", "berba-gai",
                           "berkedu-dukan", "berkelan-jutan", "ber-saing", 
                           "ber-tumpu", "ber-tindak", "berkontri-busi", 
                           "berlan", "berandan", "bersaman", 
                           "bersenjata-kelihatan", "berpan", "berbelan-", "bermaian",
                           "diempat", "berain", "berari",
                           "berbagal", # typo for 'berbagai'? based on the concordance
                           "berbasi", # typo for 'berbasi'? based on the concordance
                           "berbah", # name of place
                           "berbas", # name of place
                           "berban" # split word for 'berbanding'
                           )) %>% 
  mutate(morphind = str_replace_all(morphind, "(?<=\\+)jerit\\<v\\>", "jerit<n>"),
         morphind = replace(morphind, word_form=='menjerit-jerit', "meN+jerit<n>_VPA"),
         root_pos_morphind = ifelse(root_morphind == "jerit", "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "jerit", 
                                        str_replace(affix_morphind_wclass, "_v_", "_n_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)dagang\\<v\\>", "dagang<n>"),
         root_pos_morphind = ifelse(root_morphind == "dagang", "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "dagang", 
                                        str_replace(affix_morphind_wclass, "_v_", "_n_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)semi\\<f\\>", "semi<n>"),
         root_pos_morphind = ifelse(root_morphind == "semi", "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "semi", 
                                        str_replace(affix_morphind_wclass, "_f_", "_n_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)laku\\<a\\>", "laku<n>"),
         root_pos_morphind = ifelse(root_morphind == "laku" & word_form == "diperlaku-kan", "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "laku" & word_form == "diperlaku-kan", 
                                        str_replace(affix_morphind_wclass, "_a_", "_n_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)lebur\\<v\\>", "lebur<a>"),
         root_pos_morphind = ifelse(root_morphind == "lebur", "a", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "lebur", 
                                        str_replace(affix_morphind_wclass, "_v_", "_a_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)heran\\<v\\>", "heran<a>"),
         root_pos_morphind = ifelse(root_morphind == "heran", "a", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "heran", 
                                        str_replace(affix_morphind_wclass, "_v_", "_a_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)sehat\\<n\\>", "sehat<a>"),
         root_pos_morphind = ifelse(root_morphind == "sehat", "a", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "sehat", 
                                        str_replace(affix_morphind_wclass, "_n_", "_a_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)libur\\<n\\>", "libur<v>"),
         root_pos_morphind = ifelse(root_morphind == "libur", "v", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "libur", 
                                        str_replace(affix_morphind_wclass, "_n_", "_v_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)bagi\\<v\\>", "bagi<n>"),
         morphind = str_replace_all(morphind, "(?<=\\+)silah?kan\\<v\\>", "sila<v>+kan"),
         suff_morphind = replace(suff_morphind, root_morphind %in% c("silakan", "silahkan"), "kan"),
         affix_morphind = ifelse(root_morphind %in% c("silakan", "silahkan"), 
                                        str_replace_all(affix_morphind, "_0$", "_kan"), 
                                        affix_morphind),
         affix_morphind_wclass = ifelse(root_morphind %in% c("silakan", "silahkan"), 
                                        str_replace(affix_morphind_wclass, "_0$", "_kan"), 
                                        affix_morphind_wclass),
         root_morphind = replace(root_morphind, root_morphind %in% c("silakan", "silahkan"), "sila"),
         morphind = replace(morphind, word_form=='membagi-bagi', "meN+bagi<n>_VPA"),

         morphind = replace(morphind, word_form=='terperangkap', "ter+perangkap<n>_VSP"),
         morphind = replace(morphind, word_form=='terperangkapnya', "ter+perangkap<n>_VSP+dia<p>_PS3"),
         morphind = replace(morphind, word_form=='diperangkap', "di+perangkap<n>_VSP"),
         morphind = replace(morphind, word_form=='diperban', "di+perban<n>_VSP"),
         morphind = str_replace_all(morphind, "(?<=\\+)acuh\\<a\\>", "acuh<v>"),
         affix_morphind = replace(affix_morphind, word_form=='diperband', "di-_0"),
         affix_morphind = replace(affix_morphind, word_form=='diperangkap', "di-_0"),
         pref_morphind = replace(pref_morphind, word_form=='diperangkap', "di-"),
         pref_morphind = replace(pref_morphind, word_form=='diperban', "di-"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form=='diperangkap', "di-_n_0"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form=='diperban', "di-_n_0"),
         root_morphind = replace(root_morphind, 
                                 word_form %in% c("diperban"), "perban"),
         affix_morphind = replace(affix_morphind, 
                                  word_form %in% c("terperangkap", "terperangkapnya"), 
                                  "ter-_0"),
         pref_morphind = replace(pref_morphind, 
                                 word_form %in% c("terperangkap", "terperangkapnya"), 
                                 "ter-"),
         affix_morphind_wclass = replace(affix_morphind_wclass, 
                                         word_form %in% c("terperangkap", "terperangkapnya"), 
                                         "ter-_n_0"),
         root_morphind = replace(root_morphind, 
                                 word_form %in% c("terperangkap", "diperangkap", 
                                                  "terperangkapnya"), 
                                 "perangkap"),
         root_pos_morphind = replace(root_pos_morphind, 
                                     word_form %in% c("terperangkap", "diperangkap", "terperangkapnya"), "n"),
         
         morphind = replace(morphind, word_form=='mendamprat', "meN+damprat<n>_VSA"),
         root_morphind = replace(root_morphind, word_form=='mendamprat', "damprat"),
         morphind = replace(morphind, word_form=='mencerocos', "meN+cerocos<pre-cat>_VSA"),
         morphind = replace(morphind, word_form=='mendandani', "meN+dandan<v>+i_VSA"),
         root_morphind = replace(root_morphind, word_form=='mendandani', "dandan"),
         morphind = replace(morphind, word_form=='menepuk-nepuk', "meN+tepuk<v>_VPA"),
         morphind = replace(morphind, word_form=='membagi-bagikan', "meN+bagi<n>+kan_VPA"),
         root_pos_morphind = ifelse(root_morphind == "bagi", "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "bagi", 
                                        str_replace(affix_morphind_wclass, "_v_", "_n_"), 
                                        affix_morphind_wclass),
         
         morphind = replace(morphind, 
                            word_form %in% c("diperhatikan", "diper-hatikan"), 
                            "di+per+hati<n>+kan_VSP"),
         root_morphind = replace(root_morphind, 
                                 word_form %in% c("diperhatikan", "diper-hatikan"), 
                                 "hati"),
         root_pos_morphind = replace(root_pos_morphind, 
                                     word_form %in% c("diperhatikan", "diper-hatikan"), 
                                     "n"),
         pref_morphind = replace(pref_morphind, 
                                 word_form %in% c("diperhatikan", "diper-hatikan"), "di-+per-"),
         suff_morphind = replace(suff_morphind, 
                                 word_form %in% c("diperhatikan", "diper-hatikan"), "kan"),
         affix_morphind = replace(affix_morphind, 
                                  word_form %in% c("diperhatikan", "diper-hatikan"), "di-+per-_kan"),
         affix_morphind_wclass = replace(affix_morphind_wclass, 
                                         word_form %in% c("diperhatikan", "diper-hatikan"), 
                                         "di-+per-_n_kan"),
         
         morphind = replace(morphind, word_form == "memerhatikannya", "meN+per+hati<n>+kan_VSA"),
         pref_morphind = replace(pref_morphind, word_form == "memerhatikannya", "meN-+per-"),
         suff_morphind = replace(suff_morphind, word_form == "memerhatikannya", "kan"),
         affix_morphind = replace(affix_morphind, word_form == "memerhatikannya", "meN-+per-_kan"),
         affix_morphind_wclass = replace(affix_morphind_wclass, 
                                         word_form == "memerhatikannya", "meN-+per-_n_kan"),
         
         morphind = replace(morphind, word_form == "diempati", "di+empati<n>_VSP"),
         root_morphind = replace(root_morphind, word_form == "diempati", "empati"),
         root_pos_morphind = replace(root_pos_morphind, word_form %in% c("diempati"), "n"),
         suff_morphind = replace(suff_morphind, word_form %in% c("diempati"), "0"),
         affix_morphind = replace(affix_morphind, word_form %in% c("diempati"), "di-_0"),
         affix_morphind_wclass = replace(affix_morphind_wclass, 
                                         word_form  %in% c("diempati"), 
                                         "di-_n_0"),
         
         morphind = replace(morphind, word_form %in% c("diperbaiki", "diperbaikii"), "di+per+baik<a>+i_VSP"),
         root_morphind = replace(root_morphind, word_form %in% c("diperbaiki", "diperbaikii"), "baik"),
         root_pos_morphind = replace(root_pos_morphind, word_form %in% c("diperbaiki", "diperbaikii"), "a"),
         pref_morphind = replace(pref_morphind, word_form %in% c("diperbaiki", "diperbaikii"), "di-+per-"),
         suff_morphind = replace(suff_morphind, word_form %in% c("diperbaiki", "diperbaikii"), "i"),
         affix_morphind = replace(affix_morphind, word_form %in% c("diperbaiki", "diperbaikii"), "di-+per-_i"),
         affix_morphind_wclass = replace(affix_morphind_wclass, 
                                         word_form  %in% c("diperbaiki", "diperbaikii"), 
                                         "di-+per-_a_i"),
         
         root_morphind = replace(root_morphind, root_morphind == "disain", "desain"),
         morphind = if_else(str_detect(morphind, "(?<=\\+)disain(?=\\<n\\>)"),
                            str_replace(morphind, "(?<=\\+)disain(?=\\<n\\>)", "desain"),
                            morphind),
         
         root_morphind = replace(root_morphind, root_morphind == "ijin", "izin"),
         morphind = if_else(str_detect(morphind, "(?<=\\+)ijin(?=\\<n\\>)"),
                            str_replace(morphind, "(?<=\\+)ijin(?=\\<n\\>)", "izin"),
                            morphind),
         
         morphind = if_else(root_morphind == "bahaya", 
                            str_replace_all(morphind, "bahaya<a>", "bahaya<n>"),
                            morphind),
         
         morphind = replace(morphind, word_form == "memasak", "meN+masak<v>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("memasak"), "masak"),
         root_pos_morphind = replace(root_pos_morphind, word_form %in% c("memasak"), "v"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form == "memasak", "meN-_v_0"),
         
         verb_tagged = replace(verb_tagged, word_form %in% c("terpenuhi", "terpenihi"), TRUE),
         morphind = replace(morphind, word_form %in% c("terpenuhi", "terpenihi"), "ter+penuh<a>+i_VSP"),
         root_morphind = replace(root_morphind, word_form %in% c("terpenuhi", "terpenihi"), "penuh"),
         pref_morphind = replace(pref_morphind, word_form %in% c("terpenuhi", "terpenihi"), "ter-"),
         suff_morphind = replace(suff_morphind, word_form %in% c("terpenuhi", "terpenihi"), "i"),
         affix_morphind = replace(affix_morphind, word_form %in% c("terpenuhi", "terpenihi"), "ter-_i"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form %in% c("terpenuhi", "terpenihi"), "ter-_a_i"),
         word_form = replace(word_form, word_form == "terpenihi", "terpenuhi"),
         
         morphind = replace(morphind, word_form %in% c("terperam"), "ter+peram<pre-cat>_VSP"),
         root_morphind = replace(root_morphind, word_form %in% c("terperam"), "peram"),
         root_pos_morphind = replace(root_pos_morphind, word_form == "terperam", "pre-cat"),
         pref_morphind = replace(pref_morphind, word_form %in% c("terperam"), "ter-"),
         suff_morphind = replace(suff_morphind, word_form %in% c("terperam"), "0"),
         affix_morphind = replace(affix_morphind, word_form %in% c("terperam"), "ter-_0"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form %in% c("terperam"), "ter-_pre-cat_0"),
         
         morphind = replace(morphind, word_form %in% c("terpencarkan"), "ter+pencar<v>+kan_VSP"),
         root_morphind = replace(root_morphind, word_form %in% c("terpencarkan"), "pencar"),
         root_pos_morphind = replace(root_pos_morphind, word_form == "terpencarkan", "v"),
         pref_morphind = replace(pref_morphind, word_form %in% c("terpencarkan"), "ter-"),
         suff_morphind = replace(suff_morphind, word_form %in% c("terpencarkan"), "kan"),
         affix_morphind = replace(affix_morphind, word_form %in% c("terpencarkan"), "ter-_kan"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form %in% c("terpencarkan"), "ter-_v_kan"),
         
         morphind = if_else(root_morphind == "esa", 
                            str_replace_all(morphind, "\\<a\\>", "<c>"),
                            morphind),
         root_pos_morphind = replace(root_pos_morphind, 
                                     root_morphind == "esa", "c"),
         affix_morphind_wclass = if_else(root_morphind == "esa", 
                                         str_replace_all(affix_morphind_wclass, 
                                                         "_num_", "_c_"),
                                         affix_morphind_wclass),
         
         morphind = replace(morphind, word_form %in% c("bergeliat-geliut"), "ber+geliat-geliut<n>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("bergeliat-geliut"), "geliat-geliut"),
         
         morphind = replace(morphind, word_form %in% c("berkaca-mata"), "ber+kacamata<n>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("berkaca-mata"), "kacamata"),
         
         morphind = replace(morphind, word_form %in% c("bercocok-tanam"), "ber+cocok tanam<n>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("bercocok-tanam"), "cocok tanam"),
         root_pos_morphind = replace(root_pos_morphind, word_form == "bercocok-tanam", "n"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form %in% c("bercocok-tanam"), "ber-_n_0"),
         
         morphind = replace(morphind, word_form %in% c("berdin-ding"), "ber+dinding<n>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("berdin-ding"), "dinding"),
         
         morphind = replace(morphind, word_form %in% c("berduka-kecewa"), "ber+duka kecewa<n>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("berduka-kecewa"), "duka kecewa"),
         
         morphind = replace(morphind, word_form %in% c("berdaya-guna"), "ber+daya guna<n>_VSA"),
         morphind = str_replace_all(morphind, "dayaguna\\<n\\>", "daya guna<n>"),
         root_morphind = replace(root_morphind, word_form %in% c("berdaya-guna"), "daya guna"),
         root_morphind = replace(root_morphind, root_morphind %in% c("dayaguna"), "daya guna"),
         morphind = replace(morphind, word_form %in% c("berdaya-cipta"), "ber+daya cipta<n>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("berdaya-cipta"), "daya cipta"),
         
         morphind = replace(morphind, word_form %in% c("berdaya-kuasa"), "ber+daya kuasa<n>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("berdaya-kuasa"), "daya kuasa"),
         
         morphind = replace(morphind, word_form %in% c("terperi"), "ter+peri<n>_VSP"),
         root_morphind = replace(root_morphind, word_form %in% c("terperi"), "peri"),
         root_pos_morphind = replace(root_pos_morphind, word_form == "terperi", "n"),
         pref_morphind = replace(pref_morphind, word_form %in% c("terperi"), "ter-"),
         suff_morphind = replace(suff_morphind, word_form %in% c("terperi"), "0"),
         affix_morphind = replace(affix_morphind, word_form %in% c("terperi"), "ter-_0"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form %in% c("terperi"), "ter-_n_0"),
         
         morphind = replace(morphind, word_form %in% c("terperikan"), "ter+peri<n>+kan_VSP"),
         root_morphind = replace(root_morphind, word_form %in% c("terperikan"), "peri"),
         root_pos_morphind = replace(root_pos_morphind, word_form == "terperikan", "n"),
         pref_morphind = replace(pref_morphind, word_form %in% c("terperikan"), "ter-"),
         suff_morphind = replace(suff_morphind, word_form %in% c("terperikan"), "kan"),
         affix_morphind = replace(affix_morphind, word_form %in% c("terperikan"), "ter-_kan"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form %in% c("terperikan"), "ter-_n_kan"),
         
         root_morphind = replace(root_morphind, word_form == "beralih-rupa", "alih rupa"),
         root_morphind = replace(root_morphind, word_form == "beralih-wujud", "alih wujud"),
         root_morphind = replace(root_morphind, word_form == "beranak-cucu", "anak cucu"),
         root_morphind = replace(root_morphind, word_form == "beranak-cucu", "anak pinak"),
         
         root_morphind = replace(root_morphind, word_form == "berbaik-hati", "baik hati"),
         root_morphind = replace(root_morphind, word_form == "bercita-rasa", "cita rasa"),
         
         root_morphind = replace(root_morphind, root_morphind == "campuraduk", "campur aduk"),
         root_morphind = replace(root_morphind, word_form == "bercampur-baur", "campur baur"),
         morphind = replace(morphind, word_form == "bercampur-baur",
                            "ber+campur baur<v>_VSA"),
         morphind = replace(morphind, word_form == "bercampur-aduk",
                            "ber+campur aduk<v>_VSA"),
         morphind = replace(morphind, word_form == "tercampur-aduk",
                            "ter+campur aduk<v>_VSP"),
         root_morphind = replace(root_morphind, word_form %in% c("tercampur-aduk",
                                                                 "bercampur-aduk"), 
                                 "campur aduk"),
         morphind = if_else(root_morphind == "campur aduk",
                            str_replace_all(morphind, "campuraduk<[av]>", "campur aduk<v>"),
                            morphind),
         root_pos_morphind = replace(root_pos_morphind,
                                     root_morphind == "campur aduk" & root_pos_morphind == "a",
                                     "v"),
         affix_morphind_wclass = replace(affix_morphind_wclass,
                                     root_morphind == "campur aduk" 
                                     & affix_morphind_wclass == "meN-_a_kan",
                                     "meN-_v_kan"),
         
         root_morphind = replace(root_morphind, morphind == "ber+aneka<a>_VSADASHragam<n>_NSD", "aneka ragam"),
         root_pos_morphind = replace(root_pos_morphind, morphind == "ber+aneka<a>_VSADASHragam<n>_NSD", "n"),
         affix_morphind = replace(affix_morphind, morphind == "ber+aneka<a>_VSADASHragam<n>_NSD", "ber-_n_0"),
         root_morphind = replace(root_morphind, root_morphind == "anekaragam", "aneka ragam"),
         
         root_morphind = replace(root_morphind, morphind == "ber+aneka<a>_VSADASHmacam<n>_NSD", "aneka macam"),
         root_morphind = replace(root_morphind, morphind == "ber+aneka<a>_VSADASHrupa<n>_NSD", "aneka rupa"),
         root_pos_morphind = replace(root_pos_morphind, 
                                     morphind %in% c("ber+aneka<a>_VSADASHmacam<n>_NSD",
                                                     "ber+aneka<a>_VSADASHrupa<n>_NSD"), "n"),
         affix_morphind = replace(affix_morphind, 
                                  morphind %in% c("ber+aneka<a>_VSADASHmacam<n>_NSD",
                                                  "ber+aneka<a>_VSADASHrupa<n>_NSD"), "ber-_n_0"),
         root_morphind = replace(root_morphind, root_morphind == "anekaragam", "aneka ragam"),
         
         morphind = replace(morphind, word_form %in% c("berbasis-bank"), "ber+basis-bank<n>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("berbasis-bank"), "basis-bank"),
         
         # word_form = if_else(word_form %in% c("berbeda-bagi", "berbeda-dalam", 
         #                                      "bercahaya-mungkin", "berburu-meramu",
         #                                      "bercerita-dengan", "bercerita-entah"),
         #                     str_replace_all(word_form, "\\-[a-z]+$", ""),
         #                     word_form),
         
         morphind = replace(morphind, word_form %in% c("berazan"), "ber+azan<n>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("berazan"), "azan"),
         root_pos_morphind = replace(root_pos_morphind, word_form == "berazan", "n"),
         suff_morphind = replace(suff_morphind, word_form %in% c("berazan"), "0"),
         affix_morphind = replace(affix_morphind, word_form %in% c("berazan"), "ber-_0"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form %in% c("berazan"), "ber-_n_0"),
         
         morphind = replace(morphind, word_form %in% c("berbasa-basi"), "ber+basa basi<n>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("basabasi"), "basa basi"),
         
         morphind = replace(morphind, word_form %in% c("berbulan-madu"), "ber+bulan madu<n>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("berbulan-madu"), "bulan madu"),
         
         root_morphind = replace(root_morphind, root_morphind == "ceraiberai", "cerai berai"),
         
         morphind = replace(morphind, word_form %in% c("terperinci"), "ter+per+rinci<a>_VSP"),
         root_morphind = replace(root_morphind, word_form %in% c("terperinci"), "rinci"),
         root_pos_morphind = replace(root_pos_morphind, word_form == "terperinci", "a"),
         pref_morphind = replace(pref_morphind, word_form %in% c("terperinci"), "ter-+per-"),
         suff_morphind = replace(suff_morphind, word_form %in% c("terperinci"), "0"),
         affix_morphind = replace(affix_morphind, word_form %in% c("terperinci"), "ter-+per-_0"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form %in% c("terperinci"), "ter-+per-_a_0"),
         
         morphind = replace(morphind, word_form == "bertempik-sorak", "ber+tempik sorak<n>_VSA"),
         root_morphind = replace(root_morphind, word_form == "bertempik-sorak", "tempik sorak"),
         root_pos_morphind = replace(root_pos_morphind, word_form == "bertempik-sorak", "n"),
         pref_morphind = replace(pref_morphind, word_form == "bertempik-sorak", "ber-"),
         affix_morphind_wclass = replace(affix_morphind_wclass, 
                                         word_form == "bertempik-sorak", 
                                         "ber-_n_0"),
         affix_morphind = replace(affix_morphind, word_form == "bertempik-sorak", "ber-_0"),
         
         morphind = replace(morphind, word_form == "diperbankan", "di+per+bank<n>+kan_VSP"),
         root_morphind = replace(root_morphind, word_form == "diperbankan", "bank"),
         
         morphind = replace(morphind, word_form == "berbesan", "ber+besan<n>_VSA"),
         root_morphind = replace(root_morphind, word_form == "berbesan", "besan"),
         root_pos_morphind = replace(root_pos_morphind, word_form == "berbesan", "n"),
         pref_morphind = replace(pref_morphind, word_form == "berbesan", "ber-"),
         suff_morphind = replace(suff_morphind, word_form == "berbesan", "0"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form == "berbesan", "ber-_n_0"),
         affix_morphind = replace(affix_morphind, word_form == "berbesan", "ber-_0"),
         
         morphind = replace(morphind, word_form == "berkejap-kerjap", "ber+kejap kerjap<n>_VPA"),
         root_morphind = replace(root_morphind, word_form == "berkejap-kerjap", "kejap kerjap"),
         root_pos_morphind = replace(root_pos_morphind, word_form == "berkejap-kerjap", "n"),
         pref_morphind = replace(pref_morphind, word_form == "berkejap-kerjap", "ber-"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form == "berkejap-kerjap", "ber-_n_0"),
         affix_morphind = replace(affix_morphind, word_form == "berkejap-kerjap", "ber-_0"),
         
         morphind = replace(morphind, word_form=='diperolah', "di+per+oleh<r>_VSP"),
         morphind = replace(morphind, word_form=='memperolah', "meN+per+oleh<r>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("diperolah", "memperolah"), "oleh"),
         root_pos_morphind = replace(root_pos_morphind, word_form %in% c("diperolah", "memperolah"), "r"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form=='diperolah', "di-+per-_r_0"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form=='memperolah', "meN-+per-_r_0"),
         affix_morphind = replace(affix_morphind, word_form=='diperolah', "di-+per-_0"),
         affix_morphind = replace(affix_morphind, word_form=='memperolah', "meN-+per-_r_0"),
         
         
         morphind = if_else(word_form %in% c("meminta", "memintakan"), 
                            str_replace_all(morphind, "pinta<n>", "minta<v>"), 
                            morphind),
         root_morphind = if_else(word_form %in% c("meminta", "memintakan"),
                                 str_replace(root_morphind, "^pinta$", "minta"),
                                 root_morphind),
         root_pos_morphind = if_else(word_form %in% c("meminta", "memintakan"),
                                     str_replace(root_pos_morphind, "^n$", "v"),
                                     root_pos_morphind),
         affix_morphind_wclass = if_else(word_form %in% c("meminta", "memintakan"),
                                         str_replace(affix_morphind_wclass, "_n_", "_v_"),
                                         affix_morphind_wclass),
         
         morphind = if_else(word_form %in% c("mengecek"), 
                            str_replace_all(morphind, "kecek<n>", "cek<v>"), 
                            morphind),
         root_morphind = if_else(word_form %in% c("mengecek"),
                                 str_replace(root_morphind, "^kecek$", "cek"),
                                 root_morphind),
         root_pos_morphind = if_else(word_form %in% c("mengecek"),
                                     str_replace(root_pos_morphind, "^n$", "v"),
                                     root_pos_morphind),
         affix_morphind_wclass = if_else(word_form %in% c("mengecek"),
                                         str_replace(affix_morphind_wclass, "_n_", "_v_"),
                                         affix_morphind_wclass),
         
         morphind = if_else(word_form %in% c("mencek"), 
                            str_replace_all(morphind, "cek<n>", "cek<v>"), 
                            morphind),
         root_pos_morphind = if_else(word_form %in% c("mencek"),
                                     str_replace(root_pos_morphind, "^n$", "v"),
                                     root_pos_morphind),
         affix_morphind_wclass = if_else(word_form %in% c("mencek"),
                                         str_replace(affix_morphind_wclass, "_n_", "_v_"),
                                         affix_morphind_wclass),
         
         morphind = if_else(word_form %in% c("menanyai", "menanyakan", "menanya"), 
                            str_replace_all(morphind, "(?<=\\+)(nanya)(?=<v>)", "tanya"), 
                            morphind),
         root_morphind = if_else(word_form %in% c("menanyai", "menanyakan", "menanya"),
                                 str_replace(root_morphind, "^nanya$", "tanya"),
                                 root_morphind),
         
         morphind = if_else(word_form %in% c("menerabas"), 
                            str_replace_all(morphind, "(?<=\\+)(nerabas)(?=<v>)", "terabas"), 
                            morphind),
         root_morphind = if_else(word_form %in% c("menerabas"),
                                 str_replace(root_morphind, "^nerabas$", "terabas"),
                                 root_morphind),
         
         morphind = if_else(root_morphind == "rasa", str_replace_all(morphind, "rasa<v>", "rasa<n>"), morphind),
         root_pos_morphind = if_else(root_morphind == "rasa", "n", root_pos_morphind),
         affix_morphind_wclass = if_else(root_morphind == "rasa", 
                                         str_replace_all(affix_morphind_wclass,
                                                         "_v_", "_n_"),
                                         affix_morphind_wclass),
         
         word_form = replace(word_form, word_form=="terangin", "diterangin"),
         morphind = replace(morphind, word_form=="diterangin", "di+terang<a>+in_VSP"),
         root_morphind = replace(root_morphind, word_form=="diterangin", "terang"),
         root_pos_morphind = replace(root_pos_morphind, word_form=="diterangin", "a"),
         pref_morphind = replace(pref_morphind, word_form=="diterangin", "di-"),
         suff_morphind = replace(suff_morphind, word_form=="diterangin", "in"),
         affix_morphind_wclass = replace(affix_morphind_wclass, word_form=="diterangin", "di-_a_in"),
         affix_morphind = replace(affix_morphind, word_form=="diterangin", "di-_in"),
         dbase = replace(dbase, word_form=="diterangin", "di"),
         verb_tagged = replace(verb_tagged, word_form=="diterangin", TRUE),
         
         morphind = replace(morphind, word_form=='berjual-beli', "ber+jualbeli<n>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("berjual-beli"), "jualbeli"),
         root_pos_morphind = replace(root_pos_morphind, word_form %in% c("berjual-beli"), "n"),
         affix_morphind_wclass = replace(affix_morphind_wclass, 
                                         word_form=='berjual-beli', "ber-_n_0"),
         
         morphind = replace(morphind, word_form=='memperinci', "meN+per+rinci<a>_VSA"),
         root_morphind = replace(root_morphind, word_form %in% c("memperinci"), "rinci"),
         root_pos_morphind = replace(root_pos_morphind, word_form %in% c("memperinci"), "a"),
         affix_morphind_wclass = replace(affix_morphind_wclass, 
                                         word_form=='memperinci', "meN-+per-_a_0"),
         
         morphind = str_replace_all(morphind, "(?<=\\+)giat\\<n\\>", "giat<a>"),
         root_pos_morphind = replace(root_pos_morphind, root_morphind == "giat", "a"),
         affix_morphind_wclass = ifelse(root_morphind == "giat", 
                                        str_replace(affix_morphind_wclass, "_n_", "_a_"), 
                                        affix_morphind_wclass),
         
         morphind = replace(morphind, word_form=='dipertanggung', "di+per+tanggungjawab<n>+kan_VSP"),
         morphind = replace(morphind, word_form=='mempertanggung', "meN+per+tanggungjawab<n>+kan_VSA"),
         root_morphind = replace(root_morphind, 
                                 word_form %in% c("dipertanggung", "mempertanggung"), 
                                 "tanggungjawab"),
         suff_morphind = replace(suff_morphind, 
                                 word_form %in% c("dipertanggung", "mempertanggung"), 
                                 "kan"),
         root_pos_morphind = replace(root_pos_morphind, 
                                     word_form %in% c("dipertanggung", "mempertanggung"), "n"),
         affix_morphind_wclass = replace(affix_morphind_wclass, 
                                         word_form=='dipertanggung', "di-+per-_n_kan"),
         affix_morphind_wclass = replace(affix_morphind_wclass, 
                                         word_form=='mempertanggung', "meN-+per-_n_kan"),
         affix_morphind = replace(affix_morphind, 
                                  word_form=='dipertanggung', "di-+per-_kan"),
         affix_morphind = replace(affix_morphind, 
                                  word_form=='mempertanggung', "meN-+per-_kan"),
         
         
         morphind = replace(morphind, word_form=='diperjual', "di+per+jualbeli<n>+kan_VSP"),
         morphind = replace(morphind, word_form=='memperjual', "meN+per+jualbeli<n>+kan_VSA"),
         root_morphind = replace(root_morphind, 
                                 word_form %in% c("diperjual", "memperjual"), 
                                 "jualbeli"),
         suff_morphind = replace(suff_morphind, 
                                 word_form %in% c("diperjual", "memperjual"), 
                                 "kan"),
         root_pos_morphind = replace(root_pos_morphind, 
                                     word_form %in% c("diperjual", "memperjual"), 
                                     "n"),
         affix_morphind_wclass = replace(affix_morphind_wclass, 
                                         word_form=='diperjual', 
                                         "di-+per-_n_kan"),
         affix_morphind_wclass = replace(affix_morphind_wclass, 
                                         word_form=='memperjual', 
                                         "meN-+per-_n_kan"),
         affix_morphind = replace(affix_morphind, 
                                  word_form=='diperjual', "di-+per-_kan"),
         affix_morphind = replace(affix_morphind, 
                                  word_form=='memperjual', "meN-+per-_kan"),
         
         morphind = replace(morphind, word_form=='berbasi', "ber+basi<a>_VSA"),
         morphind = replace(morphind, word_form=='bergelar', "ber+gelar<n>_VSA"),
         morphind = replace(morphind, word_form=='melinting', "meN+linting<n>_VSA"),
         root_pos_morphind = ifelse(word_form=="berbasi", "a", root_pos_morphind),
         root_morphind = ifelse(word_form=="berbasi", "basi", root_morphind),
         root_pos_morphind = ifelse(word_form=="bergelar", "n", root_pos_morphind),
         affix_morphind_wclass = replace(affix_morphind_wclass, 
                                         word_form == "berbasi", 
                                         "ber-_a_0"),
         affix_morphind = replace(affix_morphind, 
                                  word_form == "berbasi", 
                                  "ber-_0"),
         affix_morphind_wclass = replace(affix_morphind_wclass, 
                                         word_form == "bergelar", 
                                         "ber-_n_0"),
         morphind = str_replace_all(morphind, "(?<=\\+)robek\\<v\\>", "robek<a>"),
         morphind = replace(morphind, word_form=='merobek-robek', "meN+robek<a>_VPA"),
         root_pos_morphind = ifelse(root_morphind == "robek", "a", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "robek", 
                                        str_replace(affix_morphind_wclass, "_v_", "_a_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)seberang\\<v\\>", "seberang<n>"),
         root_pos_morphind = ifelse(root_morphind == "seberang", "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "seberang", 
                                        str_replace(affix_morphind_wclass, "_v_", "_n_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)lisan\\<a\\>", "lisan<n>"),
         root_pos_morphind = ifelse(root_morphind == "lisan", "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "lisan", 
                                        str_replace(affix_morphind_wclass, "_a_", "_n_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)lenting\\<a\\>", "lenting<v>"),
         morphind = str_replace_all(morphind, "(?<=\\+)lengkung\\<n\\>", "lengkung<a>"),
         root_pos_morphind = ifelse(root_morphind == "lengkung", "a", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "lengkung", 
                                        str_replace(affix_morphind_wclass, "_n_", "_a_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)jalan\\<v\\>", "jalan<n>"),
         root_pos_morphind = ifelse(root_morphind == "jalan", "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "jalan", 
                                        str_replace(affix_morphind_wclass, "_v_", "_n_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)teriak\\<v\\>", "teriak<n>"),
         root_pos_morphind = ifelse(root_morphind == "teriak", "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "teriak", 
                                        str_replace(affix_morphind_wclass, "_v_", "_n_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)cakap\\<v\\>", "cakap<n>"),
         root_pos_morphind = ifelse(root_morphind == "cakap", "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "cakap", 
                                        str_replace(affix_morphind_wclass, "_v_", "_n_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)gumam\\<v\\>", "gumam<n>"),
         root_pos_morphind = ifelse(root_morphind == "gumam", "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "gumam", 
                                        str_replace(affix_morphind_wclass, "_v_", "_n_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)sorak\\<v\\>", "sorak<n>"),
         root_pos_morphind = ifelse(root_morphind == "sorak", "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "sorak", 
                                        str_replace(affix_morphind_wclass, "_v_", "_n_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)tahan\\<v\\>", "tahan<a>"),
         root_pos_morphind = ifelse(root_morphind == "tahan", "a", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "tahan", 
                                        str_replace(affix_morphind_wclass, "_v_", "_a_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)kisar\\<v\\>", "kisar<n>"),
         root_pos_morphind = ifelse(root_morphind == "kisar", "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "kisar", 
                                        str_replace(affix_morphind_wclass, "_v_", "_n_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)padu\\<v\\>", "padu<a>"),
         root_pos_morphind = ifelse(root_morphind == "padu", "a", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "padu", 
                                        str_replace(affix_morphind_wclass, "_v_", "_a_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "((?<=^ber\\+)tanggal\\<v\\>(?=_V)|(?<=^ter\\+)tanggal\\<v\\>(?=_V))", "tanggal<n>"),
         root_pos_morphind = ifelse(word_form %in% c("tertanggal", "bertanggal"), "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(word_form %in% c("tertanggal", "bertanggal"), 
                                        str_replace(affix_morphind_wclass, "_v_", "_n_"), 
                                        affix_morphind_wclass),
         morphind = str_replace_all(morphind, "(?<=\\+)bujang\\<f\\>", "bujang<n>"),
         root_pos_morphind = ifelse(root_morphind == "bujang", "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(root_morphind == "bujang", 
                                        str_replace(affix_morphind_wclass, "_f_", "_n_"), 
                                        affix_morphind_wclass),
         morphind = ifelse(word_form %in% c("berbikini", "bertopi", "berhelai", "berhelai-helai", "berselai", "berponi", "beruntai", "berderai", "berderailah", "berderai-derailah", "berderai-", "berderai-derai", "beresponsi", "bermigasi", "beranalogi"), str_replace_all(morphind, "(?<=^ber\\+)([^<]+)\\<.\\>\\+(i)", "\\1\\2<n>"), morphind),
         root_pos_morphind = ifelse(word_form %in% c("berbikini", "bertopi", "berhelai", "berhelai-helai", "berselai", "berponi", "beruntai", "berderai", "berderailah", "berderai-derailah", "berderai-", "berderai-derai", "beresponsi", "bermigasi", "beranalogi"), "n", root_pos_morphind),
         affix_morphind_wclass = ifelse(word_form %in% c("berbikini", "bertopi", "berhelai", "berhelai-helai", "berselai", "berponi", "beruntai", "berderai", "berderailah", "berderai-derailah", "berderai-", "berderai-derai", "beresponsi", "bermigasi", "beranalogi"), 
                                        str_replace(affix_morphind_wclass, "_[a-z]_i$", "_n_0"), 
                                        affix_morphind_wclass),
         affix_morphind = ifelse(word_form %in% c("berbikini", "bertopi", "berhelai", "berhelai-helai", "berselai", "berponi", "beruntai", "berderai", "berderailah", "berderai-derailah", "berderai-", "berderai-derai", "beresponsi", "bermigasi", "beranalogi"), 
                                        str_replace(affix_morphind, "_i$", "_0"), 
                                        affix_morphind),
         root_morphind = ifelse(word_form %in% c("berbikini", "bertopi", "berhelai", "berhelai-helai", "berselai", "berponi", "beruntai", "berderai", "berderailah", "berderai-derailah", "berderai-", "berderai-derai", "beresponsi", "bermigasi", "beranalogi"), str_extract(morphind, "([^<+]+?)(?=<)"), root_morphind))

more_editing <- read_tsv('more_editing_database.txt') %>% 
  filter(dbase=='me')
verbs <- verbs %>% 
  left_join(more_editing) %>% 
  mutate(morphind = ifelse(word_form %in% more_editing$word_form & suff_morphind=="0" & str_detect(morphind, "X--"), 
                           paste("meN+", ROOT2, "<", root_pos_morphind, ">_VSA", sep = ""), 
                           morphind), 
         morphind = ifelse(word_form %in% more_editing$word_form & suff_morphind!="0" & str_detect(morphind, "X--"), 
                           paste("meN+", ROOT2, "<", root_pos_morphind, ">+", suff_morphind, "_VSA", sep = ""), 
                           morphind), 
         root_morphind = ifelse(word_form %in% more_editing$word_form, ROOT2, root_morphind)) %>% 
  select(-ROOT2)